How Criticker fits into my personal movie recommendations...

[td888]

Hi Quicky, very nice! You do all this in Excel?

[Quicky]

Hi Quicky, very nice! You do all this in Excel?

The hardcore calculations of the weights and final scores for each movie are done in IDL, the one programming language I'm rather fluent in. The presentation is done in Excel, yes. I'm rather happy with all the improvements in Excel 2007. Lots of nice graphical stuff to play with .

[td888]

Great work. Maybe you should create some kind of 'plug-in' site for criticker, which generates these kind of graphs, statistics, etc... for a criticker user. I don't know if there any other criticker-users interested, but I sure am.

I don't know how 'portable' your code is, but it should be possible. With a small 'automated' site and some google-adsense it should be self-supporting. Anyway, I got some idea's to help you out. If you're interested, just PM me.

[Philip]

Updated list for AFlickering: http://thequicky.net/files/aflickering- ... tions4.pdf
Recommendations list for philip: http://thequicky.net/files/philip-qmdb- ... ations.pdf

Thanks a lot! Now I've got yet another reason for feeling bad about not having seen the godfather

The keywords and the other stats are just as interesting as the rankings, even though I can't really see how "studio-logo-segues-into-film" could possibly affect my opinion about a movie.

I like what I see on the esthetic side too, keep up the good work! And once again, thanks.

[Quicky]

[...] even though I can't really see how "studio-logo-segues-into-film" could possibly affect my opinion about a movie.

Here's a quote from Wikipedia that should shed some light on the issue of weird keywords coming up .

The conventional dictum that "correlation does not imply causation" means that correlation cannot be validly used to infer a causal relationship between the variables. This dictum should not be taken to mean that correlations cannot indicate causal relations. However, the causes underlying the correlation, if any, may be indirect and unknown. Consequently, establishing a correlation between two variables is not a sufficient condition to establish a causal relationship (in either direction).

This basically says that just because it appears that you like those movies that have 'shot-in-the-head' as a keyword, doesn't mean there's a causal relationship between a movie having that keyword and you liking the movie. It might just be a coincidence. Of course, if everything about these correlations would be coincidence, then there would be no use for these recommendations. However, I trust that at least some of these keywords, as well as the other parameters do have a causal relationship with you liking those movies .

I like what I see on the aesthetic side too, keep up the good work! And once again, thanks.

Thank!

[paulofilmo]

Is the bell-shaped tier system retrieving different results, or is it just for presentation?

[Quicky]

Is the bell-shaped tier system retrieving different results, or is just for presentation?

It's just for the presentation of a few ratings. I use it to help me rate other movies, by checking between which movies the movie I want to rate fits in best.

[paulofilmo]

That's put me on a whole new train of thought. I'll have a go at putting it into words later.

Are there any improvements of your system that you're working on? Any flaws that you're aware of? Ways in which you think it could be improved?

[Quicky]

Are there any improvements of your system that you're working on? Any flaws that you're aware of? Ways in which you think it could be improved?

As of now, I'm primarily trying to optimize the code to make it faster, which is basically just to make it faster for me to use. The results won't (at least shouldn't!) be influenced by this. One flaw I'm aware of is that the optimization of the parameters is not perfect, since it doesn't always catch the absolute maximum. I've been looking for ways to fix this, but have so far been unable to find a better algorithm. I'm not aware of any other flaws. It's hard to determine the flaws and merits of each parameter, so it's hard to find out whether the system would work better if you add or remove a certain parameter. One other flaw is that I don't have every movie on Criticker or IMDB in my database. However I'm confident that I have all the most popular and the highest rated movies on IMDB. Furthermore, every time I make the list for a new user, I add all the movies that that user has seen and are not in my database yet.

Oh and of course... being a perfectionist in heart and soul I keep trying to make the .pdf files prettier and more useful .

[Philip]

I discovered something while going through the movies to put them on my remember list. I don't think "Grave of the Fireflies" (http://www.criticker.com/?f=4316) should be in the humor section. "This movie makes me want to kill myself" (80 pts), "One of the most depressing and beautiful movies I've ever seen." (90 pts), "Too sad for words..." (81 pts) ...

I haven't seen the movie, so maybe humor is a sub category, but I don't really think it's a laugh out loud kinda film.